Have you looked for an album on Spotify on its release day and been met with only no internet connection messages? Or waited in a Ticketmaster queue only to be kicked out and forced to log back in? These instances are called outages or crashes and are typical occurrences on low-end websites, but they are starting to be increasingly noticeable as sites like Spotify, Ticketmaster, and Google fall victim to these glitches. What causes crashes, why are they happening more often, and what can be done to prevent them?
What causes a crash?
At midnight on June 30, thousands of fans opened their Spotify apps hoping to listen to Olivia Rodrgio’s new single “Vampire” and Lil Uzi Vert’s third album “Pink Tape” were dismayed when they were unable to find the music. Those lucky enough to find it were met with three dot loading screens and no internet connection signs. Spotify crashed due to the high volume of people visiting the site and trying to access the same music. Users that weren’t trying to listen to the new songs may have not even noticed that Spotify had crashed. Crashes are caused by a mismatch between traffic levels and capacity of the website’s infrastructure. When several people make system requests, especially the same system requests the processing capacity of the servers are exceeded. There is an assumption that bigger sites should be able to handle a large amount of visitors, however the amount of people accessing the site only grows as the dominance of the site increases. It is also important to note that not all system requests hold the same weight. Requests to view the home page or play previously downloaded music require much less of the app, but requests to listen to new music that is being played by thousands of other people at the same time takes a lot for the software to process.
Why are sites created to handle large numbers of people crashing?
Many remember the Ticketmaster fiasco of 2022 when Eras Tour tickets went on sale whether or not they were personally affected by it. Ticketmaster accidentally sent out too many presale codes to eager fans. All of these eager fans plus fans who had not received presale codes and bots tried to enter the queue when the tickets went on sale. The site began almost immediately to buckle under the demands from millions of users. Visitors were often kicked out of the queue, alerted the sale was no longer available when it was, and were let into the sale space only to discover there were no longer tickets. This occurrence has continued to happen each time a big artist puts their tickets on sale through Ticketmaster. Ticketmaster was built to be a ticket seller that works effortlessly so why are sales continuously aggravating hopeful buyers? The answer is actually quite simple. Fanbases are at a much larger scale than they were about 10 years ago. Through social media and artist campaigns a lot more people are introduced to or aware of musical artists. Everyone wants to see their concerts either because they love the artist or because they want to have something to brag about to their friends. This desire has created an expansive market that is very attractive to bots and off market sellers. About one third of the 3.5 billion visitors who tried to access the Eras Tour presale were bots. What once was a small manageable group of prospective buyers cascaded into a expansive wave of visitors all hoping to get tickets for different reasons.
So what can sites do to optimize the quality and accessibility of their services?
Even Google, one of the world’s most used sites can’t escape the demand of fanbases. When Taylor Swift created a vault puzzle to be solved on Google, it crashed only a couple hours after it was released after millions of people tried to access it. The demand will likely only continue to grow. Adaptations must be made in each site in order to optimize the quality and accessibility the services they provide. Constant proactive monitoring will alerts IT to failures in real time and provide insight into traffic level and current performance levels. Sites that don’t already content distribution network, should do so as soon as possible as it responds to browser requests for your static content, home pages and product images. This frees up servers to focus on dynamic content like checkout processes. It might be counter-intuitive but sites that receive high demand on a regular basis would be wise to downgrade the user experience limiting the site to the simplest and most essential function to make those functions as efficient as possible. This would be a great first step for sites like Ticketmaster. A final option would be to introduce autoscaling that adjusts the functions and user experience based on the demand day to day. For example an site may expand its server capacity while reducing avaible functions. Autoscaling would be extremely helpful for Spotify, the site could expand its server capacity on album release days to improve the accessibility of the music at the time it is released.